- AngelsRound
- Posts
- Bounty ✨
Bounty ✨
AI agent marketplace for completed, verified work.

Spotlight
What if businesses paid for finished work instead of software?
Quick Pitch: Bounty is building a marketplace where businesses hire AI agents to complete tasks, with payment released after verification.

The Problem
Subscription Costs: Businesses pay for software regardless of usage.
Execution Overhead: AI tools still require setup, prompting, and oversight.
Outsourcing Friction: Hiring freelancers and verifying work takes time.

Why It Matters
AI adoption is widespread, but scaling it remains difficult.
McKinsey's 2025 State of AI report found that 88% of organizations use AI in at least one function, yet only 7% have fully scaled it.
As AI agents handle more business tasks, companies could increasingly purchase completed services instead of managing multiple software tools.
Snapshot
Industry: AI Agents / Future of Work
Headquarters: New York, NY
Founded: 2026
Backed By: a16z Speedrun
Funding: Raising $2M (Seed)
Business Model: 20% fee on individual tasks; 30% on premium
Traction: $1.13M annualized GMV; 10K+ users; 2K+ completed tasks
Target Customers: Founders, startups, and small businesses
Founder’s Edge
Aashna Doshi, Co-Founder, CEO: Former Google AI infrastructure engineer, Georgia Tech ML master's graduate, with 200K+ social followers.
Shivam Patel, Co-Founder, CTO: Former Barclays engineer who helped scale a startup to $2M ARR.
Playing Field
Freelance Marketplaces: Upwork, Fiverr, and Toptal connect businesses with human service providers but require customers to manage hiring and quality.
AI Software: SaaS tools provide capabilities but leave execution to customers.
Bounty's Edge: Customers pay for verified results without managing tools or freelancers.
Analysis
Bulls Case 📈
Outcome based pricing could capture SaaS and freelance spending.
Automated verification could lower fulfillment costs.
Early GMV suggests demand for AI delivered services.
More specialized agents could expand task coverage.
Bears Case 📉
Complex tasks may be costly to verify.
Low switching costs limit defensibility.
Upwork and Fiverr have established buyers and growing AI offerings.
Retention and unit economics remain unproven.

Verdict
Bounty faces an interesting tradeoff. Simple tasks are easier to verify but face pricing pressure, while complex work commands higher fees but costs more to validate. Its opportunity lies in repeatable, valuable tasks where verification remains inexpensive. Whether Bounty can own that segment before established marketplaces do will determine its long term advantage.
The Startup Pulse
Another happening week in startup funding. Three signals from this week:
Investors are putting serious money behind automating AI research itself
Inference hardware is attracting billion dollar valuations before commercial production
AI agents are creating demand for evaluation, monitoring and enterprise infrastructure
Mirendil: Raised $1B at a $5B valuation, led by Kleiner Perkins and Clearlake. Building AI agents that automate research experiments.
Mercor: Landed $750M at a $20B valuation, backed by Benchmark, Felicis and Nvidia. Connects AI labs with experts for model training and evaluation.
Fractile: Raised $720M at a $7B valuation, backed by Thrive, Founders Fund and Lightspeed. Developing AI inference chips to reduce memory bottlenecks.
Strategic breakdowns of how the biggest AI companies actually work — their business models, their moats, their vulnerabilities. This week: Rockets are now SpaceX’s smallest business
Written by Ashher